B2B Export Sales Workspace

A bilingual, local-first workspace that helps export sales teams turn an English buyer message into a traceable customer record, quotation, and follow-up task.

What problem it solves

Export sales work often moves between inboxes, spreadsheets, chat tools, and individual memory. That makes it easy to:

• quote before specifications, quantity, or destination are confirmed;
• confuse Gross Margin with Markup;
• re-enter the same customer and inquiry facts on multiple pages;
• lose the relationship between an inquiry, quotation, and follow-up;
• miss overdue work or prioritize customers by incomplete profile data alone.

This application creates one small operating workspace for those decisions. SQLite stores the long-term business relationships; Streamlit Session State is used only for temporary page-to-page context.

Core workflow

Analyze Inquiry → Prepare Quotation → Follow Up

The application preserves the customer, inquiry, matched product, quotation, and follow-up relationships across the workflow. It does not attempt to replace an ERP, customs platform, or order-execution system. Optional enhanced inquiry analysis is supported, but the complete core workflow runs without an API key.

Live Demo

Open the public Streamlit workspace or use the Live Demo badge above. The deployed application uses fictional data and works in local-rules mode without an API key.

Streamlit Community Cloud may restart the application and reinitialize its local SQLite database. Treat the online workspace as a resettable demonstration, not durable storage, and do not enter confidential customer information.

Screenshots

Public bilingual workspace

End-to-end workflow

Decision-focused analytics

How to use it

1. Open Analyze Inquiry and load the fictional demo inquiry or paste an English RFQ.
2. Review the request summary, missing information, risks, recommended questions, product match, and editable professional English reply.
3. Link an existing customer or create a new customer, then save the inquiry.
4. Open Create Quotation. Customer, inquiry, product, quantity, specification, destination, and database IDs are inherited automatically.
5. Enter costs, exchange rate, pricing method, terms, and validity, then save the quotation.
6. Create a linked follow-up. The customer, inquiry, quotation, stage, and recommended follow-up date are inherited.

The language control in the upper-right switches the interface between English and Simplified Chinese without clearing the current workflow context or form inputs. Customer-facing suggested replies remain professional English.

Core capabilities

Area	What is included
Inquiry	Demo inquiry, structured summary, graded information gaps, risk categories, recommended questions, editable English reply
Customers	Independent Data Completeness and commercial Lead Quality, filters, next action, editing, import/export, business timeline
Products	Compact specification, MOQ, cost, packaging, sample lead time, production lead time, and match-usage management
Quotations	EXW, FOB, CIF, DDP, CNY-to-USD conversion, Gross Margin or Markup, per-quotation exchange rate, Excel export
Follow-ups	Overdue/today/future task queue, customer/stage/priority/date filters, Inquiry and Quotation references
Analytics	Customer count, Lead Quality distribution, overdue work, conversion rates, funnel, source, and country distributions
Settings	Editable default exchange rate, concise analysis status, fictional demo-data ensure/reset controls

Professional customer portfolio

Data Completeness measures whether the customer record is usable. Lead Quality is separate and uses commercial value: purchase potential, buying progress, persisted inquiry/quotation evidence, and engagement.

Follow-up task queue

Tasks are grouped into overdue, today, and future work. Each persisted task shows its customer, stage, priority, Inquiry ID, and Quotation ID.

Unmatched product handling

An older inquiry without a matched product can be rematched, assigned manually, or used to calculate an unmatched draft. Formal saving requires explicit product verification.

Quotation logic

All costs are entered in CNY. The exchange-rate definition is:

1 USD = X CNY
USD quote = CNY quote ÷ X

The workspace setting supplies a default exchange rate. Each quotation can override it without changing the default.

Gross Margin — default

Selling price = Cost ÷ (1 - Gross Margin rate)

Gross Margin measures profit as a percentage of selling price.

Markup — alternative

Selling price = Cost × (1 + Markup rate)

Markup measures profit as a percentage of cost. A 25% Gross Margin and a 25% Markup do not produce the same selling price, so the selected method is labeled and persisted.

Estimated DDP results retain an explicit warning because destination-country tariff, tax, customs, and final-delivery assumptions require verification.

Architecture

flowchart LR
    U["Export salesperson"] --> P["Streamlit pages"]
    P --> W["Workflow services"]
    W --> R["Parameterized repositories"]
    R --> DB[("SQLite")]
    W --> Q["Quotation engine"]
    W --> I["Local inquiry rules"]
    I --> O["Optional enhanced analysis"]
    W --> X["Excel import/export"]

Loading

• pages/ owns presentation and user interaction.
• services/ owns inquiry, customer-quality, matching, quotation, follow-up, analytics, and workflow rules.
• database/ owns initialization, forward migrations, repositories, and fictional demo data.
• locales/ contains stable English and Simplified Chinese translation keys.
• components/ contains the approved visual system and workflow context.

Database relationships

erDiagram
    CUSTOMERS ||--o{ INQUIRIES : receives
    PRODUCTS ||--o{ INQUIRIES : matched_to
    CUSTOMERS ||--o{ QUOTATIONS : receives
    INQUIRIES ||--o{ QUOTATIONS : produces
    PRODUCTS ||--o{ QUOTATIONS : quoted_as
    CUSTOMERS ||--o{ FOLLOW_UPS : has
    INQUIRIES ||--o{ FOLLOW_UPS : references
    QUOTATIONS ||--o{ FOLLOW_UPS : triggers
    CUSTOMERS ||--o{ ACTIVITIES : timeline
    INQUIRIES ||--o{ ACTIVITIES : timeline
    QUOTATIONS ||--o{ ACTIVITIES : timeline

Loading

Core persisted links:

• inquiries.customer_id
• inquiries.matched_product_id
• quotations.customer_id
• quotations.inquiry_id
• quotations.product_id
• follow_ups.customer_id
• follow_ups.inquiry_id
• follow_ups.quotation_id
• optional activities.inquiry_id and activities.quotation_id

Migrations are forward-only and idempotent. They preserve existing records, avoid duplicate columns/indexes, and are tested with SQLite integrity and foreign-key checks.

Fictional demo data

All bundled companies, people, domains, phone numbers, products, inquiries, costs, quotations, follow-ups, and business outcomes are fictional.

The clean demo contains:

• 20 customers
• 10 inquiries
• 8 quotations
• 10 follow-ups
• 6 products

Ensure demo data adds missing bundled records. Reset fictional demo data replaces only bundled demo records and preserves user-created records. Browser QA data is created in temporary databases and is not part of the seed.

Installation

Python 3.11 or later is recommended.

git clone https://github.com/hql7-luo/b2b-export-sales-intelligence.git
cd b2b-export-sales-intelligence

python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -r requirements.txt

cp .env.example .env
python -m database.init_db

Windows activation:

.venv\Scripts\activate

OPENAI_API_KEY is optional. Leave it empty to use local inquiry rules.

Run

source .venv/bin/activate
streamlit run app.py

Open http://localhost:8501.

To use a separate database:

DATABASE_PATH=data/another-workspace.db streamlit run app.py

A new database initializes automatically and receives fictional demo data on its first application run.

Test

python -m pytest

Optional coverage:

python -m pytest --cov=services --cov=database --cov-report=term-missing

Tests cover pricing formulas, inquiry analysis, repository safety, migrations, workflow relationships, translation keys, demo reset behavior, release files, and application smoke rendering.

Deployment

GitHub Actions

.github/workflows/tests.yml installs requirements.txt with Python 3.11 and runs the complete pytest suite on pushes and pull requests.

Streamlit Community Cloud

The repository includes the root requirements.txt and .streamlit/config.toml expected by Streamlit Community Cloud.

The public demonstration is available at:

b2b-export-sales-intelligence-qtonipxh5e4bwnfst2a5zj.streamlit.app

To deploy another instance:

1. Select app.py as the entrypoint.
2. Select Python 3.11 in Advanced settings.
3. Leave OPENAI_API_KEY empty to use the complete local-rules workflow.
4. Add secrets only when optional enhanced analysis is intentionally enabled.

References: Streamlit deployment, dependencies, and secrets.

The online Demo uses Streamlit-local SQLite and may reset or reinitialize when the container restarts or the app is redeployed. It is suitable for a demonstration, not durable multi-user production data. Use a managed database and authentication before handling real customer information.

Colab demo

Open the Google Colab demo to run seven code cells directly from the GitHub main branch. It demonstrates customer scoring, inquiry analysis, Gross Margin versus Markup, quotation outputs, and decision-focused business charts using only fictional data. No API key is required.

Limitations

• This is a portfolio-grade single-workspace application, not a multi-tenant production system.
• SQLite has no built-in user authentication and is plaintext at rest.
• Streamlit Community Cloud storage may be ephemeral.
• DDP is an estimate until destination-country tariff, tax, customs, and final-delivery inputs are verified.
• Product matching is explainable keyword matching, not a product feasibility guarantee.
• Local inquiry rules are deterministic and do not replace salesperson review.
• External analysis, when enabled, sends inquiry text to the configured provider; confidential data requires authorization and an appropriate data policy.
• ERP, order execution, customs-data feeds, automated email/calendar actions, and advanced predictive models are intentionally out of scope.

Project structure

.
├── app.py
├── pages/
│   ├── inquiry_analyzer.py
│   ├── customers.py
│   ├── quotation_calculator.py
│   ├── follow_up_tracker.py
│   ├── products.py
│   ├── dashboard.py
│   └── settings.py
├── services/
│   ├── inquiry_analyzer.py
│   ├── inquiry_brief.py
│   ├── customer_intelligence.py
│   ├── product_match.py
│   ├── quotation.py
│   ├── workflow.py
│   ├── followup.py
│   └── dashboard.py
├── database/
│   ├── connection.py
│   ├── init_db.py
│   ├── migrations.py
│   ├── repository.py
│   └── seed_data.py
├── components/
├── locales/
├── tests/
├── docs/screenshots/
├── notebooks/export_sales_intelligence_demo.ipynb
├── .github/workflows/tests.yml
├── .streamlit/
├── LICENSE
├── requirements.txt
└── README.md

Resume description

• Built a bilingual Streamlit and SQLite export-sales workspace that persists the complete Inquiry → Quotation → Follow-up relationship and customer timeline.
• Implemented a tested quotation engine for EXW, FOB, CIF, DDP, Gross Margin, Markup, per-quotation exchange rates, and Excel output.
• Designed decision-focused customer quality, follow-up queue, product reference, and sales analytics modules using Python, SQL, Pandas, and Plotly.